J. Phys. Chem. B 2009, 113, 4131–4140
4131
Infrared Spectroscopy and Optical Constants of Porous Amorphous Solid Water† Franc¸ois Cholette,‡ Tykhon Zubkov,§,⊥ R. Scott Smith,§ Zdenek Dohna´lek,§ Bruce D. Kay,*,§ and Patrick Ayotte*,‡ Fundamental and Computational Sciences Directorate, Pacific Northwest National Laboratory, P.O. Box 999, Mail Stop K8-88, Richland, Washington 99352, and De´partement de chimie, UniVersite´ de Sherbrooke, 2500 BouleVard UniVersite´, Sherbrooke, Que´bec, Canada J1K 2R1 ReceiVed: July 29, 2008; ReVised Manuscript ReceiVed: September 17, 2008
Reflection-absorption infrared spectra (RAIRS) of amorphous solid water (ASW) films grown at 20 K on a Pt(111) substrate at various angles (θBeam ) 0-85°) using a molecular beam are reported. They display complex features arising from the interplay between refraction, absorption within the sample, and interference effects between the multiple reflections at the film-substrate and film-vacuum interfaces. Using a simple classical optics model based on Fresnel equations, we obtain optical constants [i.e., n(ω) and k(ω)] for porous ASW in the 1000-4000 cm-1 (10-2.5 µm) range. The behavior of the optical properties of ASW in the intramolecular OH stretching region with increasing θBeam is shown to be strongly correlated with its decreasing density and increasing surface area. A direct comparison between the RAIRS and calculated vibrational spectra shows a large difference (∼200 cm-1) in the position of the coupled H-bonded intramolecular OH stretching vibrations spectral feature. Moreover, this band shifts in opposite directions with increasing θBeam in RAIRS and vibrational spectra demonstrating RAIRS spectra cannot be interpreted straightforwardly as vibrational spectra due to severe optical distortions from refraction and interference effects. I. Introduction Amorphous solid water (ASW) is an important substrate and matrix for the chemistry of the interstellar medium (ISM).1-3 For example, it is believed to be an important constituent of the complex molecular solids, or ices, that form comets, the frosty mantle of interstellar dust grains in cold and dense molecular clouds that shrouds protostellar and protoplanetary regions, as well as at the surface of cold outer solar system objects.1-5 Depending on a variety of factors including the local temperature, incident H2O flux, incidence angle, and kinetic energy, crystalline or amorphous ices formed by vapor condensation can have various amounts of porosity. Therefore, physical data obtained using molecular beam methods6-8 are crucial to accurately model the growth and properties of these molecular solids.9,10 Interstellar ices may also be contaminated by other condensable material either during their growth or by subsequent adsorption and uptake within their porous structure. Clearly, the interaction of gas phase molecules with dust grains in the ISM will depend greatly on the morphology and microstructure of their ASW mantle.11-14 Increasingly detailed models of the structure, morphology, and composition of the icy coating on interstellar grains are continuously being developed based on the interpretation of spectroscopic observations of the ISM. Consequently, accurate optical constants obtained through laboratory studies of condensed water substances as astrophysical ices surrogates15-17 are essential for an accurate interpretation of spectroscopic data, for example † Part of the special section “Aqueous Solutions and Their Interfaces”. * To whom correspondence should be addressed. (P.A.) E-mail:
[email protected]. Phone: (819) 821-7889. Fax (819) 8218017. (B.D.K.) E-mail:
[email protected]. Phone: (509) 371-6143. Fax: (509) 371-6145. ‡ Universite ´ de Sherbrooke. § Pacific Northwest National Laboratory. ⊥ Present address: Department of Chemistry, Ball State University, Muncie, IN 47306.
those from the Infrared Space Telescope (ISO),18 or from planetary probes such as those recently provided from the surface of Enceladus by Cassini.19 Significant challenges exist in the characterization of vapordeposited condensed water substances in the laboratory which account for most of the disagreement in the literature about their properties. Appropriate methods need to be developed for controlled and reproducible growth and the nondestructive characterization of these fragile nanoporous materials. We have shown that the structure and morphology of ASW films can be best controlled using molecular beam deposition techniques.20-27 Specifically, rational synthesis of nanoporous ASW films of specifically controlled crystallinity,25 density,26 porosity,21,22 and surface area23,24 can be achieved by simply controlling the incidence angle,20 flux, and kinetic energy27 of gas phase H2O molecules impinging on a cold substrate. Optical interferometry26 and reflection-absorption infrared spectroscopy (RAIRS)25,27 as well as adsorption of weakly physisorbed molecules (N2, CH4, Ar, CO, O2)20-24 have been used to probe the structure and morphology of molecular beam-deposited nanoporous ASW. Further, the complex coupled adsorption, desorption, transport and uptake kinetics of these weakly physisorbed species within nanoporous ASW films have been studied in great detail.23,24 In this work, we used a H2O molecular beam to prepare (nominally) 100 monolayers (ML) coverage (dense ASW equivalent) films at 20 K on a Pt(111) single crystal substrate as a function of deposition incidence angle. These growth conditions yield highly reproducible morphologies for which quantitative evaluation of the sample’s macroscopic density26 and surface area23,24 were previously reported using optical interferometry and nitrogen adsorption, respectively. RAIRS characterization was performed on these nanoscopic films and their optical constants were extracted using a simple classical optics model.28,29 This procedure is required in order to untangle
10.1021/jp806738a CCC: $40.75 2009 American Chemical Society Published on Web 11/08/2008
4132 J. Phys. Chem. B, Vol. 113, No. 13, 2009
Cholette et al.
Figure 1. Schematic of the experimental setup used to synthesize and perform spectroscopic characterization of nanoscopic porous ASW films. Films were grown by applying a constant exposure corresponding to a (nominal) coverage of 100 ML (dense ASW equivalent) using an effusive molecular beam [average kinetic energy 5 kJ/mol, flux ) (0.87 ML/sec) · cos θBeam] at an incident angle θBeam with respect to the normal of the Pt(111) single crystal substrate at T ) 20 K. These conditions yield a macroscopic density FASW(θBeam) and a thickness of h ) 100 ML · [FASW(θBeam ) 0°)/FASW(θBeam)]. Reflection absorption infrared spectroscopy (RAIRS) is performed by focusing unpolarized light from a FTIR spectrometer at θIR ) 82° incidence angle onto the substrate. Experimental absorbance spectra are reported using reflectivities from the film/substrate system [i.e., I(ω)] and from the clean Pt(111) substrate [i.e., Io(ω)]: A(ω) ) -log[I(ω)/Io(ω)]. Optical constants for nanoporous ASW are obtained using a simple classical optics model (see text) that describes, using Fresnel coefficients, the reflectivity of infrared radiation by infinite parallel interfaces between vacuum, a nanoporous ASW film of absolute thickness h and frequency-dependent complex index of refraction n˜ASW(ω), condensed on an optically thick platinum substrate with frequency-dependent complex index of refraction n˜Pt(ω).
the trivial refraction and interference effects, inherent to RAIRS spectroscopy,30-32 from chemically relevant absorption within the nanoscopic films. The amplitude of the spectroscopic signatures from coupled H-bonded and dangling OH intramolecular stretching vibrational modes in the optical parameters of nanoporous ASW are observed to correlate strongly with its density and surface area, respectively. II. Experimental Section All experiments were performed in a previously described apparatus,21,22 which is a stainless steel ultrahigh vacuum (UHV) chamber with a base pressure of 18 MΩ/cm) using a quasieffusive, room temperature (5 kJ/ mol average kinetic energy), molecular beam. The water vapor (2 Torr) was expanded through a 1 mm diameter aperture, and the resulting beam was triply differentially pumped. The H2O beam flux was ∼0.87 ML/s for normal incidence and scaled with deposition angle, θBeam, as cos θBeam. The beam was well collimated but broad enough to cover the entire sample at all
θBeam as schematically shown in Figure 1. Under these conditions, H2O molecules stick to ice and Pt(111) with essentially unit probability for all θBeam.6,7 A H2O exposure that produced the characteristic TPD spectrum of a saturated first monolayer coverage of H2O on Pt(111) was defined as 1 ML H2O.34 Porous ASW films having a constant coverage of 100 ML were deposited at various θBeam by adjusting the dose times according to the incidence angle-dependent molecular beam flux yielding a constant exposure of 100 ML dense ASW equivalents (i.e., dose time at θBeam given by 115s/cos θBeam). This coverage represents a compromise between the long dose times that are required at the most grazing incidences and the improving signal-to-noise ratio resulting from the spectroscopy and analysis of thicker films. As the films’ porosity develops from increasing initial surface roughness, very thin films (h < 10 ML) do not have the fully developed porosity that characterizes films hundreds of ML thick.27 Therefore, the complex sample morphological features and resulting physical properties which results in severe spectral distortions reported here do not apply to films 1-10 ML thick. RAIRS was performed in situ at grazing angle, θIR ) (82 ( 1)°, using unpolarized light from a commercial FTIR spectrometer (Bruker Equinox 55). An external optical bench described previously25 included mirrors, KBr windows and lenses for focusing the unpolarized infrared beam onto the sample and collecting the reflected light onto a narrow-band HgCdTe detector (MCT, model 313). Infrared spectra were acquired at 4 cm-1 resolution by averaging 2000 scans (∼8 min) at 20 K. Absorbance spectra are reported as A(ω) ) -log [I(ω)/Io(ω)] using the reflectivity of the film/substrate system [I(ω)] and that of the clean Pt(111) substrate [Io(ω)], both recorded at 20 K across the ω ) 800-6500 cm-1 effective spectral range of the spectrometer. RAIRS spectra of nanoscopic films display complex features arising from refraction, from interference effects (arising from the multiple reflections at the film-vacuum and the film-substrate interfaces), as well as from absorption within the film as shown schematically in Figure 1.30-32 Analysis of the RAIRS spectra was performed using a simple classical optics model. Specifically, the model describes the interaction of the incident IR beam at the various interfaces using the polarization- and frequencydependent Fresnel reflection and transmission coefficients. These are used to calculate the substrate and film/substrate system unpolarized reflectivities, R(ω) ) 1/2 [Rs(ω) + Rp(ω)], where
Porous Amorphous Solid Water
Figure 2. Selected raw absorbance spectra recorded at T ) 20 K and θIR ) 82° for (nominally) 100 ML (dense ASW equivalent) coverage nanoporous ASW films grown on Pt(111) at T ) 20 K with an effusive H2O molecular beam at θBeam ) 0, 40, 55, 75, 80, and 85°. Inset: expanded view of the dangling OH stretching vibrational features assigned to three- (i.e., DAA) and two- (i.e., DA) coordinated water molecules.
Rs and Rp represent the s- and p-polarized reflectivities.28,29 Optical constants for platinum from the literature [i.e., the real and imaginary part of its frequency-dependent complex index of refraction, n˜Pt(ω) ) nPt(ω) + ikPt(ω) ]35 are used to calculate the substrate unpolarized reflectivity, RPt(ω). Fresnel spectra (i.e., A(ω) ) -log [RASW/Pt(ω)/RPt(ω)]) are then least-squares fitted to experimental absorbance spectra while iteratively adjusting the ASW films optical constants [i.e., n˜ASW(ω) ) nASW(ω) + ikASW(ω)] until convergence. The real and imaginary parts of the complex index of refraction of ASW throughout the 800-6500 cm-1 range were constrained using the KramersKronig (KK) relation.28,29 The effect of the incompleteness of the KK transform (arising from the limited spectral coverage at ω < 800 cm-1 due to the transmission cutoff of KBr and the narrow band MCT detector) was verified to be smaller than the experimental noise. ASW films were too thin to display absorption features due to the weaker overtones and combination bands with ω > 3800 cm-1. The model input parameters are the platinum substrate optical constants,35 the sample optical index of refraction (i.e., n∞)26 and absolute thickness (i.e., h), the angle of incidence of the infrared beam [θIR ) (82 ( 1)°] and a frequency-independent scalar offset to compensate for thermal and electronic detector drift. As discussed in details below, the only adjustable parameter in our model is the absolute sample thickness. III. Results RAIRS spectra were recorded for ASW films having a (nominal) coverage of 100 ML (dense ASW equivalent) on Pt(111) at 20 K for θBeam ranging from normal incidence (0°) to grazing incidence (85°) in 5° increments. These conditions are known to yield dense ASW films for θBeam < 40° and ASW films of increasing porosity for θBeam > 40°.20-27 Figure 2 displays representative RAIRS spectra for dense (0 and 40°), moderately porous (55°), and highly porous (75, 80, and 85°) films over the 800-6500 cm-1 available spectral range of the FTIR spectrometer. All spectra display absorption features centered near 900 cm-1 (intermolecular librations band), 1600 cm-1 (intramolecular HOH bending band), 2200 cm-1 (bending/ libration combination band), and 3400 cm-1 (intramolecular OH stretching band) assigned to coupled vibrational modes of water molecules in the disordered H-bonding network of ASW.36,37
J. Phys. Chem. B, Vol. 113, No. 13, 2009 4133 These positive absorbance bands are superimposed on a smoothly decreasing negative absorbance baseline with increasing wavenumbers which arises from destructive interferences (between the multiple reflections at the film-vacuum and film-substrate interfaces) within the ASW film.28,29 The greater negative absorbance with increasing wavenumbers arises from two effects: the greater relative optical thickness of the ASW film at shorter wavelengths (yielding larger phase shifts and increasingly destructive interferences) as well as the lower reflectivity of bare platinum for wavenumbers greater than 3500 cm-1. In the inset to Figure 2, we show a magnified view of the dangling OH bonds intramolecular stretching vibrations spectral features peaking at 3695 cm-1 (single donor-double acceptor or DAA bonding configurations) and 3720 cm-1 (single donor-single acceptor or DA bonding configurations).38-40 These spectral features are assigned to under-coordinated water molecules that have one of their intramolecular OH bonds protruding in vacuum (i.e., dangling or free OH) either from thefilmsurface,orfromwithinthefilms’porousmicrostructure.38-40 Surprisingly, only subtle changes in the RAIRS vibrational spectra are observed despite the dramatic decrease in density between films dosed at near normal incidence [FASW(θBeam ) 0-40°) > 0.90 g/cm3] and highly porous samples dosed at glancing angle [FASW(θBeam > 70°) < 0.60 g/cm3].26 First, the smooth baseline negative slope increases slightly with increasing θBeam. Second, the amplitude of all absorption features, including that of dangling OH bonds (Figure 2, inset), increases continuously with increasing θBeam. Finally, there is a slight red-shift in the large absorption feature centered near 3400 cm-1. This red-shift arises from a pronounced increase in absorbance on the lower wavenumbers edge of this broadband which is assigned to coupled H-bonded intramolecular OH stretching vibrations of H2O molecules within the bulk of ASW.36,37 One could naively interpret the increasing intensity of the dangling bonds spectral features as arising from the greater abundance of molecules with DAA and DA bonding topologies according to the films increasing porosity and surface area at larger θBeam.23,24,27 As a greater total number of water molecules lie at the film-vacuum interface in a porous ASW film comparatively to a dense ASW film, one expects a greater intensity for the dangling OH stretching vibrations spectral features with increasing θBeam. In these single donor (i.e., DAA and DA) bonding topologies, under-coordinated surface-bound molecules are expected to have their other intramolecular OH bond engaged in a stronger H-bond with a neighboring H2O molecule. Such single-donor bonding topologies are known to yield more red-shifted Hbonded OH stretching frequencies along with larger corresponding oscillator strengths than those of fully coordinated (i.e., double donor-double acceptor or DDAA) H2O molecules in the bulk of ASW.41,42 This correlation in the spectral shifts and related intensities of the dangling and H-bonded intramolecular OH stretching vibrations spectral features and greater relative abundance of surface-bound H2O molecules (DAA and DA) compared to fully coordinated molecules (DDAA) in the bulk of ASW could thus explain the evolution of the RAIRS spectra in the OH stretching range with increasing θBeam (i.e., increasing porosity and surface area).23,24,26 However, the interpretation of such small changes in the infrared spectra of porous ASW is not straightforward as RAIRS spectra are strongly distorted by trivial optical effects. Specifically, they display a complex interplay between refraction and interference effects (between the multiple reflections at the filmvacuum and film-substrate interfaces) as well as absorption
4134 J. Phys. Chem. B, Vol. 113, No. 13, 2009
Figure 3. Fresnel model input parameters. Optical indices of refraction, n∞, for molecular beam grown ASW films as a function of θBeam were taken from ref 26 and fitted to an analytical expression (continuous line, left ordinate). Values used in the model are indicated by open circles. Porous ASW films densities (obtained from their n∞ using Lorentz-Lorenz relation) were then used to calculate absolute film thicknesses, h, as a function of θBeam (dashed line, right ordinate). However, absolute film thicknesses were rather obtained spectroscopically using infrared interferometry (see text) yielding the empirical values shown as open squares (1 ML ) 3.67 Å) which show very good agreement with those calculated using the film densities.
within the film.30-32 For porous ASW films with complex morphologies, the details of the nanoscale textures are not probed at IR wavelengths (λ > 3 µm). Thus, at the macroscopic level, evolution in their morphology with increasing θBeam is only reflected in modulations of their bulk properties namely, as a dramatic decrease in their density and optical indices of refraction, n∞.26 Access to the molecular-level distortions in the H-bonded network of the porous nanostructures of ASW is nevertheless encoded in their RAIRS spectra. However, in order to properly interpret these subtle spectroscopic changes, one needs to untangle the chemically relevant absorption of radiation, through vibrational excitations within the film, from the trivial optical effects inherent to RAIRS. As we will demonstrate below, severe distortions by optical effects in RAIRS spectra complicate their analysis in terms of vibrational spectroscopy to an extent that invalidates the simple qualitative interpretation given above. We used a simple classical optics model to describe the interaction of the IR beam with nanoscopic ASW films condensed on an optically thick platinum substrate in vacuum. While all ASW films have a constant 100 ML (dense ASW equivalent) coverage (i.e., a surface number density of ∼1.25 × 1017 H2O molecules/cm2),43 porous films deposited at large θBeam have greater absolute physical thicknesses, h, and lower optical indices of refraction, n∞, owing to their lower density, FASW(θBeam).26 As both h and n∞ depend on the macroscopic ASW density,26 their values are coupled through the LorentzLorenz (LL) relation and must thus be self-consistently determined. The n∞ for molecular beam deposited nanoporous ASW on Pt(111) at 20 K as a function of θBeam has been measured previously using optical interferometry (i.e., at λ ) 6328Å).26 An analytical expression was obtained by fitting the experimental values and is reported as a continuous line in Figure 3. This expression was used to interpolate n∞ as a function of θBeam, and the values used in our simulations are reported as open circles in Figure 3. Using the LL relation, FASW was determined from their n∞ as a function of θBeam.26 We could therefore evaluate the porous ASW films absolute physical thickness as a function of θBeam: h ) 100 ML*[FASW(θBeam ) 0°)/
Cholette et al.
Figure 4. Bottom panel: typical Fresnel model results (continuous line) show excellent agreement with experimental data (open circles, only 1 in 20 data points are shown for clarity) across the whole mid-IR range (ω ) 800-6000 cm-1) for both dense (θBeam ) 10°, black) and highly porous (θBeam ) 80°, gray) ASW films. The inset shows details of the dangling OH stretching vibrations spectral features. Top panel: frequency-dependent real [i.e., nASW: (ω)] and imaginary [i.e., kASW: (ω)] parts of the complex indices of refraction, n˜ASW(ω) for dense (θBeam ) 10°, black) and highly porous (θBeam ) 80°, gray) ASW films.
FASW(θBeam)] where the thickness of 1 ML ASW is defined as 3.67 Å. These values are reported as a dashed line in Figure 3. In our model, however, we treated h as an adjustable parameter. This introduced some flexibility to allow for small fluctuations in the actual molecular beam flux and the resulting film thickness at the various θBeam. The absolute ASW film thickness was determined using IR interferometry. Specifically, as absorption of IR radiation by thin ASW films is negligible above 4000 cm-1, we adjusted h in the model in order to fit the experimental absorbance in this transparent portion of the RAIRS spectra. The optimal h for films deposited at θBeam as obtained through IR interferometry are reported as open squares in Figure 3. These empirical values are in very good agreement with the expected film thicknesses calculated using FASW (dashed line). Using the values of n∞ and h reported in Figure 3, and θIR ) 82°, we modeled the RAIRS spectra of nanoporous ASW films and obtained their frequency-dependent optical constants [n˜ASW(ω) ) nASW(ω) + ikASW(ω)] at the various θBeam. Representative results from our simulations for dense (θBeam ) 10°, black) and highly porous (θBeam ) 80°, gray) ASW films are reported in Figure 4. In the bottom panel, we compare experimental data (open symbols, 1 in 20 data points are shown for clarity) with simulation results (continuous lines) over the 800-6000 cm-1 range. The inset shows a magnified view of the dangling OH stretching vibrations spectral region. The frequency-dependent real [i.e., nASW(ω)] and imaginary [i.e., kASW(ω)] parts of their complex indices of refraction are compared in the upper panel of Figure 4. Despite the great similarities in the RAIRS spectra of dense and highly porous ASW films (Figure 4, bottom panel), there are several qualitative differences in their optical properties. At first glance, the most striking difference is the much smaller magnitude of the kASW(ω) displayed by the porous film compared to the dense film (Figure 4, upper panel). This can be explained by the much smaller density (i.e., FASW ∼ 0.31
Porous Amorphous Solid Water versus 0.94 g/cm3),26 and resulting greater physical thickness (h ) 300 ML versus 100 ML), of the porous film compared to those of the dense ASW film. Clearly thus, in order for porous ASW films to yield a comparable absorbance to that of dense ASW films having the same (nominal) coverage (Figure 4, bottom panel), thicker porous films must display smaller absorption coefficients [i.e., kASW(ω)] than those displayed by the thinner dense films. The smaller nASW(ω) values displayed by porous ASW (Figure 4, upper panel) are simply due its smaller density and optical index of refraction (see Figure 3). Another important effect is the qualitatively different peak shapes and positions in the RAIRS spectra for dense and porous ASW (Figure 4, bottom panel) compared to the respective spectral features in their kASW(ω) (Figure 4, upper panel). For example, the amplitude of kASW(ω) for the H-bonded intramolecular OH stretching vibrational band decreases and its peak position shifts to larger wavenumbers, from kmax ) 0.67 and ω ) 3233 cm-1 for dense ASW (black), to kmax ) 0.18 and ω ) 3267 cm-1 for highly porous ASW (gray). This behavior is exactly opposite to that observed in their absorbance spectra where porous ASW (gray) showed a more intense and redshifted (i.e., maximum absorbance of 0.114 at ω ) 3413 cm-1) H-bonded intramolecular OH stretching feature compared to that of dense ASW (black) (i.e., maximum absorbance of 0.106 at ω ) 3434 cm-1). The optical effects responsible for this notable distortion of the absorption features in RAIRS are well-known and arise from refraction and anomalous dispersion [i.e., decreasing n(ω) with increasing wavenumbers] near strong peaks in k(ω). Specifically, the relative phases of the electronic and nuclear polarizations within the absorbing media change sign across an absorption feature giving rise to the characteristic “first derivative” shape of n(ω) in the vicinity of a vibrational resonance.44 In strongly absorbing ASW, this causes nASW(ω) to become smaller than 1.0 on the blue edge of the coupled H-bonded intramolecular OH stretching vibrations absorption band (i.e., in the 3350-3550 cm-1 range of Figure 4, top panel). This results in a dramatic increase in the optical path (i.e., negative refraction) in this region of the RAIRS spectra for glancing incidence radiation [i.e., θIR ) 82°]. Consequently, absorbance features in the RAIRS spectra of nanoscopic ASW films peak at higher wavenumbers than the corresponding peaks in their kASW(ω), an effect that is most pronounced in the intense coupled H-bonded intramolecular OH stretching vibrations spectral range. Furthermore, this effect is much stronger in dense ASW films which display a more intense peak in their kASW(ω), therefore causing a more pronounced anomalous dispersion for nASW(ω) in the vicinity of their intense H-bonded OH stretching absorption feature compared to porous ASW films (Figure 4, top panel). Refraction is thus very strongly coupled to the absorption of infrared radiation within the films and therefore explains the complex relationship between the frequencydependent optical constants and the RAIRS spectra for ASW films having different densities. Unfortunately, these effects are present even at the smallest film thicknesses and surprisingly, interference effects are noticeable even for increments in film thickness as small as 1 ML for very thin ASW films on platinum. However, complications due to interference effects can be largely circumvented by using substrates which are better reflectors of IR radiation than platinum: gold,31 aluminum,31 or copper.32 These observations demonstrate the necessity of untangling trivial optical effects from RAIRS spectra for a proper interpretation of their qualitative features.
J. Phys. Chem. B, Vol. 113, No. 13, 2009 4135
Figure 5. Representative optical constants for molecular beam grown nanoporous ASW films deposited at θBeam) 15, 40, 55, 65, 75, 80, and 85° at T ) 20 K on Pt(111). The inset shows details of kASW(ω) in the dangling OH bonds intramolecular stretching vibrations spectral range. The amplitude of both dangling OH stretching bands is observed to increase up to 65-75° and to decrease with further increases in θBeam whereas all other absorption bands show a continuously decreasing amplitude with increasing θBeam.
Finally, Figure 5 displays the optical constants for nanoporous ASW, deposited at selected representative values of θBeam, obtained from a detailed analysis of their RAIRS spectra (i.e., Figure 4) using our Fresnel model. Optical constants throughout the 1000-4000 cm-1 range for all deposition incidence angles are available as Supporting Information. In the bottom traces, one observes a monotonic decrease in the amplitude of kASW(ω) with increasing θBeam, an effect that appears to correlate with the larger h displayed by ASW films with greater porosity (i.e., Figure 3) as discussed previously. On the contrary, the inset to Figure 5 shows that the amplitude of the dangling OH stretching vibrational features in kASW(ω) grows up to θBeam ) 65-75°, and then decreases for greater incidence angles. As is seen in the top traces, nASW(ω) decreases continuously across most of the spectral range with increasing θBeam. This effect is mainly due to the decreasing n∞ (and density) with increasing θBeam (i.e., Figure 3).26 The opposite behavior is observed in the vicinities of the “dip” spanning the 3350-3550 cm-1 range [where nASW(ω) 60°.23,24,27 Therefore, the decreasing surface number density of dense ASW columns resulting from the increasing pore volume would result in a decreasing volumetric surface area with increasing θBeam (i.e., Figure 6b). This morphological model may thus explain why the integrated kDangling and the ASW films volumetric surface area reach a maximum around θBeam ) 65-70° and then decline at the most grazing angles (i.e., Figure 6b) while the films density decreases continuously (i.e., Figure 6a) with increasing θBeam. Finally, Figure 6c presents the total integrated absorbance [R(ω)*h ) 4πωk(ω)*h] from dangling OH bonds, calculated using optical constants for nanoporous ASW films (Figure 5) and integrated over their dangling bonds OH stretching spectral features, as a function of θBeam (total d-OH, full diamonds left ordinate). As this property describes the total absorbance from all the dangling OH bonds within the nanoporous ASW film of absolute physical thickness h, this property should be compared with their total effective surface area, σ.23,24 This latter
J. Phys. Chem. B, Vol. 113, No. 13, 2009 4137 is presented in Figure 6c [in units of the saturated nitrogen coverage (i.e., ML N2) for a (nominally) 100 ML (dense ASW equivalent) coverage nanoporous ASW film] as a function of θBeam (σ, open squares - right ordinate).23,24 A fair correlation is observed between the total integrated absorbance from all the nanoporous ASW films dangling bonds OH stretching spectral features and their total effective surface area.23,24 In nitrogen (and other weakly physisorbed species) adsorption,20-27 the contribution from capillary condensation (which may allow for multilayer adsorption in the intermediate sized pores formed at θBeam ) 45-70° incidence through the Kelvin effect) is difficult to distinguish from physisorption onto the extended interconnected nanoporous network internal surface area within the ASW films. One may therefore argue that a spectroscopic determination of the surface area for nanoporous ASW films provides a more reliable evaluation of this property. Therefore, the decreasing apparent volumetric (Figure 6b, open squares) and total (Figure 6c, open squares) effective surface area for θBeam > 65-75° observed by nitrogen adsorption may stem from a decreasing contribution from capillary condensation within the very large pores created by molecular beam deposition at these most grazing incidence angles.20-27 Finally, one concludes that different but complementary properties of nanoporous ASW films are probed by N2 physisorption and IR spectroscopic experiments which explains the somewhat poorer correlation between surface area measurements performed by these two methods (i.e., Figure 6b and 6c) than that observed between the integrated kBonded and the density of ASW (i.e., Figure 6a). B. Comparison with previously published optical parameters for ASW. The optical constants reported here for ASW should be compared to the most widely used spectroscopic data by the astrophysical community, namely those of Hudgins et al.17 and those of Le´ger et al.15 These latter authors used backfilling of their vacuum chamber to grow ASW films (0.5-1 µm thick at a maximum rate of 20 µm/hr) by vapor deposition on a high purity silicon substrate maintained at T ) 77 K. Hudgins et al.17 used a tube doser to grow their sample (0.41 µm thick at a rate of 2.6 µm/hr) on a CsI substrate at T ) 10 K (and then heated them to T ) 40, 80, 100, 120, and finally 140 K). Both these preparation methods may have yielded some adsorption on both the front and rear surfaces of the transparent substrate. The absolute sample thickness was determined by optical interferometry but neither author reported the density of their samples. Transmission absorption spectra were acquired at normal incidence (θIR ) 0°) and at the deposition temperature (as well as at the subsequent annealing temperatures by Hudgins et al.).17 Le´ger et al.15 used n∞ ) 1.31, while Hudgins et al.17 used n∞ ) 1.32, in KK transforms in order to derive the optical constants of their ASW films. The optical parameters reported here for dense ASW (i.e., θBeam)0-30°) agree quite well with those reported by Le´ger et al.:15 our values for kASW(ω) are within -3 to +7% of theirs in the H-bonded OH stretching range (i.e., 3060-3410 cm-1), within -15 to 0% for the HOH bending range (i.e., 1500-1700 cm-1), and within 0 to +15% for the combination band range (i.e., 2100-2570 cm-1). Unfortunately, our data are too noisy below 1000 cm-1 to allow a meaningful comparison in the intermolecular librations range. Given the deposition temperature and flux used by Leger et al. to grow their samples (i.e., T ) 77 K and ∼15 ML/sec, respectively),15 a relatively dense ASW film should have formed on their substrate.20-27 This may explain the similarities between the optical constants they reported and those obtained here for dense ASW (i.e., at T )
4138 J. Phys. Chem. B, Vol. 113, No. 13, 2009 20 K for θBeam ) 0-30° and a maximum flux of ∼0.87 ML/ sec) despite the fact that they used a large n∞ ) 1.31 in the KK transform (see Figure 3). Furthermore, in experiments to be reported elsewhere, only a very weak temperature dependence was observed between T ) 20 and 120 K in the RAIRS spectra of dense ASW which further validates the similarities between their optical constants obtained at T ) 77 K and those reported here for dense ASW at T ) 20 K. Finally, the nASW(ω) obtained here agree to within -5.5 to +2% across the whole 1000-4000 cm-1 range with those reported by Leger et al. (including their ∼2% systematic overestimation in n∞).15 A comparison with the optical constants reported by Hudgins et al.17 is more difficult because of important differences in the preparation method (i.e., substrate temperature of 10 K and flux near 2 ML/sec), resulting in uncertainties over the sample morphology as well as the high n∞ ) 1.32 used in the KK transform. At this much lower temperature, their films may have been initially porous. Indeed, the optical constants they reported17 for the “as grown” film at T ) 10 K matches most closely those obtained here at θBeam ) 60° and T ) 20 K (i.e., ∼25% porous or FASW ∼ 0.75 g/cm3). The initial porosity of their sample is also supported by the very large dangling OH stretching vibrations spectral feature in their kASW(ω) that peaks at ω ∼ 3665 cm-1.17 This very red-shifted dangling OH stretching frequency could be evidence of contaminants being adsorbed on their sample. For example, N2-covered ASW films yield a dangling OH peak centered at ω ∼ 3672 cm-1.23,24 The presence of internal porosity and of surface contamination may thus have affected the evaluation of the optical constants of their “as grown” sample at T ) 10 K. Upon heating to T ) 40, 80, and 100 K, such an initially porous films would have densified resulting in a loss of intensity in the dangling OH stretching vibrations spectral feature and concomitantly, an increase in the coupled H-bonded intramolecular OH stretching vibrations spectral feature in kASW(ω).27 This is precisely what is observed in the data of Hudgins et al.17 suggesting that what they reported as temperature-dependent optical constants are rather strongly affected by morphological relaxation of an initially porous sample upon thermal annealing.27 This could have been easily verified by cooling the sample to lower temperature after annealing in order to verify whether these spectroscopic changes were reversible. Finally, the optical constants they reported for ASW at T ) 100 and 120 K are still not consistent with those obtained here for fully dense ASW (θBeam ) 0-30°) relying again on our observations of a very weak temperature dependence of the absorbance spectra of dense ASW. While kASW(ω) at these higher temperatures do not display any spectral signature from dangling OH bonds, thereby suggesting they may be dense, they resemble most those obtained here for θBeam ) 40° at T ) 20 K (i.e., 4% porous or FASW ∼ 0.9 g/cm3).26 We believe uncertainties in sample morphologies and possible contamination issues render any further comparisons impossible. C. Spectroscopic Characterization of Nanoporous ASW Films H-Bonded Network: RAIRS Is Not a Pure Vibrational Spectroscopy. Using the optical constants obtained with our Fresnel model (i.e., Figure 5), we calculated the Beer-Lambert absorptivity, R(ω) ) 4πωk(ω), and the absorbance spectra, R(ω)*h ) -log [ I(ω)/Io(ω) ], for nanoporous films having a (nominal) coverage of 100 ML (dense ASW equivalent) and report their intramolecular OH stretching spectral region in Figure 7a. These (theoretical) absorbance spectra can be interpreted in terms of the vibrational spectra for free-standing nanoporous ASW films that would be observed in normal incidence transmission absorption infrared spectroscopy [i.e.,
Cholette et al.
Figure 7. (a) Theoretical normal incidence transmission absorbance spectra modeled using the nanoporous ASW empirical optical constants (i.e., Figure 5). These spectra represent the infrared vibrational spectra expected from free-standing (nominally) 100 ML (dense ASW equivalent) coverage nanoporous ASW films (without the contributions from reflection losses and interference effects). Spectra are constructed from empirical optical constants by multiplying the Beer-Lambert absorptivity, R(ω) ) 4πωk(ω) by the absolute film thickness, h: A(ω) ) R(ω)*h ) -log[I(ω)/Io(ω)]. (b) The raw experimental absorbance RAIRS spectra from Figure 2 are reproduced for comparison. Insets show details of the dangling bonds intramolecular OH stretching vibrations spectral range.
θIR ) 0°] if one could avoid the complexities arising from trivial optical effects (i.e., refraction and reflection losses as well as interferences arising from multiple reflections at the film-vacuum interfaces). The corresponding RAIRS absorbance spectra are reproduced in Figure 7b for comparison. While the amplitude of the dangling OH bands shows a qualitatively similar continuous increase with increasing θBeam, as well as identical peak positions, in both transmission (Figure 7a, inset) and RAIRS (Figure 7b, inset) spectra, their amplitude is almost twice as large in the latter. This rather similar behavior for the dangling OH bonds spectral features in both RAIRS and vibrational spectra is to be contrasted with that of the H-bonded OH bonds spectral feature. One observes that the very intense band assigned to coupled H-bonded OH stretching vibrations in the vibrational spectra of nanoporous ASW (Figure 7a) blue shifts only slightly, and that its intensity decreases only slightly, with increasing θBeam despite the dramatic decrease in ASW density (i.e., relative density falling from 100% down to below 20%; Figure 6a). Interestingly, one notes that this behavior is exactly opposite to what is observed in the RAIRS spectra (Figure 7b): the large H-bonded OH stretching band red shifts, and its intensity increases with increasing θBeam. One also immediately observes the conspicuous ∼200 cm-1 blue shift of the absorbance maxima in the RAIRS spectra (Figure 7b) compared to those of the corresponding vibrational spectra (Figure 7a). As discussed previously, the optical distortions inherent to RAIRS are most pronounced in the large H-bonded OH stretching spectral range of ASW due to the large anomalous dispersion effects, and related negative refraction. These distortions are so severe that they invalidate the simple qualitative preliminary interpretation proposed earlier based on the evolution of the intramolecular (H-bonded and dangling) OH stretching vibrations spectral features with increasing θBeam observed
Porous Amorphous Solid Water in RAIRS spectra (Figure 2). Therefore, this demonstrates that RAIRS spectra cannot be straightforwardly interpreted as pure vibrational spectra. The small blue shift and corresponding small intensity decrease of the H-bonded OH stretching band observed in the vibrational spectra of nanoporous ASW films (Figure 7a) as a function of increasing θBeam indicate that the H-bonded network of highly porous ASW is only slightly weaker and more strained than that of dense ASW. This conclusion follows those drawn by Devlin, Buch and collaborators55 to explain the continuous blue shift of the H-bonded OH stretching band observed in the vibrational spectra of ice nanoparticles of decreasing sizes. Such small changes in the vibrational spectra of nanoporous ASW films that display a dramatic decrease in density and increase in surface area with increasing θBeam therefore support the hypothesis that they are formed by nanoscopic domains (i.e., filaments or columns) of dense ASW separated by voids.20-27 On the other hand, the amplitude of the spectral features from dangling OH bonds intramolecular stretching vibrations in the vibrational spectra increases continuously with increasing porosity and saturates at large θBeam (see also Figure 6c), similar to the behavior reported in the raw absorbance spectra (Figure 2). Our spectroscopic determination of molecular beam deposited nanoporous ASW films total surface area thus also supports the suggestion from ballistic deposition simulations21,22 that it increases rapidly for increasing θBeam from 30 to 60°, but then levels off at more grazing incidence. Spectroscopic observations reported herein are thus consistent with the proposed morphological model suggesting the nanoscale texture of molecular beam deposited porous ASW is analogous to that of other more refractory materials45-54 and which consists of columns of increasing length but similar cross sectional area yielding a relatively constant surface area for θBeam > 70°, but a continuously decreasing density with increasing θBeam. Finally, as the number density of H2O molecules is constant for all the nanoporous ASW films whose (theoretical) vibrational spectra are reported in Figure 7a, one expects that the gain in oscillator strength from surface-bound H2O molecules (Figure 6c) should come at the expense of that from H2O molecules in the bulk of ASW as the films’ porosity and surface area increase. The effect of such a conversion of bulk-like molecules to surface-like molecules with increasing surface area in nanoporous ASW films should result in a decrease in the amplitude of the H-bonded intramolecular OH stretching vibrations absorbance spectral feature as a function of θBeam concomitantly with the increase in the amplitude of the dangling OH intramolecular stretching vibrations spectral features in Figure 7a. Surprisingly, the 10-fold increase in the integrated absorbance from dangling OH bonds (Figure 6c) between dense ASW and the most porous films does not result in a measurable decrease in the integrated absorbance of the coupled H-bonded intramolecular OH stretching band (data not shown). This can be explained by the ca. 100-fold greater (average) oscillator strength of H-bonded OH stretching vibrations relative to those of dangling OH stretching vibrations. Therefore, despite a 20-fold increase in the surface area of nanoporous ASW films as a function of increasing θBeam (Figure 6c), and the concomitant 10-fold decrease in their density (Figure 6a), the small (